variant-calling/workflow.wdl

import "./tasks/mapping.wdl" as mapping
import "./tasks/Metrics.wdl" as Metrics
import "./tasks/Dedup.wdl" as Dedup
import "./tasks/deduped_Metrics.wdl" as deduped_Metrics
import "./tasks/Realigner.wdl" as Realigner
import "./tasks/BQSR.wdl" as BQSR
import "./tasks/Haplotyper.wdl" as Haplotyper
import "./tasks/corealigner.wdl" as corealigner
import "./tasks/TNseq.wdl" as TNseq
import "./tasks/TNscope.wdl" as TNscope
import "./tasks/somatic.wdl" as somatic
import "./tasks/processSomatic.wdl" as processSomatic
import "./tasks/somaticFilter.wdl" as somaticFilter
import "./tasks/annovar.wdl" as annovar
import "./tasks/vcf2maf.wdl" as vcf2maf


workflow {{ project_name }} {
    
    String sample_id
    File tumor_fastq_1
    File tumor_fastq_2
    File normal_fastq_1
    File normal_fastq_2

    String SENTIEON_INSTALL_DIR
    String SENTIEON_LICENSE
    String sentieon_docker
    String varscan_docker
    String annovar_docker
    String maftools_docker
    
    File ref_dir
    String fasta
    File dbmills_dir
    String db_mills
    File dbsnp_dir
    String dbsnp
    File germline_resource
    File? regions
    File database
    String disk_size
    String cluster_config

    Boolean set_pon
    File? pon_vcf
    File? cosmic_dir
    String? cosmic_vcf

    Boolean germline
    Boolean tnseq
    Boolean tnscope
    Boolean varscan


    call mapping.mapping as tumor_mapping {
        input: 
        group=sample_id + '_tumor',
        sample=sample_id + '_tumor',
        fastq_1=tumor_fastq_1,
        fastq_2=tumor_fastq_2,
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        pl="ILLUMINAL",
        fasta=fasta,
        ref_dir=ref_dir,
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
    }

    call Metrics.Metrics as tumor_Metrics {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        fasta=fasta,
        ref_dir=ref_dir,
        sorted_bam=tumor_mapping.sorted_bam,
        sorted_bam_index=tumor_mapping.sorted_bam_index,
        sample=sample_id + '_tumor',
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
    }

    call Dedup.Dedup as tumor_Dedup {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        sorted_bam=tumor_mapping.sorted_bam,
        sorted_bam_index=tumor_mapping.sorted_bam_index,
        sample=sample_id + '_tumor',
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
    }

    call deduped_Metrics.deduped_Metrics as tumor_deduped_Metrics {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        fasta=fasta,
        ref_dir=ref_dir,
        Dedup_bam=tumor_Dedup.Dedup_bam,
        Dedup_bam_index=tumor_Dedup.Dedup_bam_index,
        sample=sample_id + '_tumor',
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
    }

    call Realigner.Realigner as tumor_Realigner {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        fasta=fasta,
        ref_dir=ref_dir,
        Dedup_bam=tumor_Dedup.Dedup_bam,
        Dedup_bam_index=tumor_Dedup.Dedup_bam_index,
        db_mills=db_mills,
        dbmills_dir=dbmills_dir,
        sample=sample_id + '_tumor',
        docker=sentieon_docker,
        disk_size=disk_size,
        regions=regions,
        cluster_config=cluster_config
    }

    call BQSR.BQSR as tumor_BQSR {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        fasta=fasta,
        ref_dir=ref_dir,
        realigned_bam=tumor_Realigner.realigner_bam,
        realigned_bam_index=tumor_Realigner.realigner_bam_index,
        db_mills=db_mills,
        dbmills_dir=dbmills_dir,
        dbsnp=dbsnp,
        dbsnp_dir=dbsnp_dir,
        sample=sample_id + '_tumor',
        regions=regions,
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
    }

    call mapping.mapping as normal_mapping {
        input: 
        group=sample_id + '_normal',
        sample=sample_id + '_normal',
        fastq_1=normal_fastq_1,
        fastq_2=normal_fastq_2,
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        pl="ILLUMINAL",
        fasta=fasta,
        ref_dir=ref_dir,
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
    }

    call Metrics.Metrics as normal_Metrics {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        fasta=fasta,
        ref_dir=ref_dir,
        sorted_bam=normal_mapping.sorted_bam,
        sorted_bam_index=normal_mapping.sorted_bam_index,
        sample=sample_id + '_normal',
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
    }

    call Dedup.Dedup as normal_Dedup {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        sorted_bam=normal_mapping.sorted_bam,
        sorted_bam_index=normal_mapping.sorted_bam_index,
        sample=sample_id + '_normal',
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
    }

    call deduped_Metrics.deduped_Metrics as normal_deduped_Metrics {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        fasta=fasta,
        ref_dir=ref_dir,
        Dedup_bam=normal_Dedup.Dedup_bam,
        Dedup_bam_index=normal_Dedup.Dedup_bam_index,
        sample=sample_id + '_normal',
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
    }

    call Realigner.Realigner as normal_Realigner {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        fasta=fasta,
        ref_dir=ref_dir,
        Dedup_bam=normal_Dedup.Dedup_bam,
        Dedup_bam_index=normal_Dedup.Dedup_bam_index,
        db_mills=db_mills,
        dbmills_dir=dbmills_dir,
        sample=sample_id + '_normal',
        docker=sentieon_docker,
        disk_size=disk_size,
        regions=regions,
        cluster_config=cluster_config
    }

    call BQSR.BQSR as normal_BQSR {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        fasta=fasta,
        ref_dir=ref_dir,
        realigned_bam=normal_Realigner.realigner_bam,
        realigned_bam_index=normal_Realigner.realigner_bam_index,
        db_mills=db_mills,
        dbmills_dir=dbmills_dir,
        dbsnp=dbsnp,
        dbsnp_dir=dbsnp_dir,
        sample=sample_id + '_normal',
        regions=regions,
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
    }

    if (germline) {
        call Haplotyper.Haplotyper as Haplotyper {
            input:
            SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
            SENTIEON_LICENSE=SENTIEON_LICENSE,
            sample=sample_id + '_normal',
            fasta=fasta,
            ref_dir=ref_dir,
            recaled_bam=normal_BQSR.recaled_bam,
            recaled_bam_index=normal_BQSR.recaled_bam_index,
            dbsnp=dbsnp,
            dbsnp_dir=dbsnp_dir,
            regions=regions,
            docker=sentieon_docker,
            disk_size=disk_size,
            cluster_config=cluster_config
        }
    }
    
    call corealigner.corealigner as corealigner {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        fasta=fasta,
        ref_dir=ref_dir,
        sample=sample_id,
        docker=sentieon_docker,
        db_mills=db_mills,
        dbmills_dir=dbmills_dir,
        dbsnp=dbsnp,
        dbsnp_dir=dbsnp_dir,
        tumor_recaled_bam=tumor_BQSR.recaled_bam,
        tumor_recaled_bam_index=tumor_BQSR.recaled_bam_index,
        normal_recaled_bam=normal_BQSR.recaled_bam,
        normal_recaled_bam_index=normal_BQSR.recaled_bam_index,
        disk_size=disk_size,
        cluster_config=cluster_config
    }
    
    if (tnseq) {
        call TNseq.TNseq as TNseq {
            input:
            SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
            SENTIEON_LICENSE=SENTIEON_LICENSE,
            fasta=fasta,
            ref_dir=ref_dir,
            regions=regions,
            corealigner_bam=corealigner.corealigner_bam,
            corealigner_bam_index=corealigner.corealigner_bam_index,
            dbsnp=dbsnp,
            dbsnp_dir=dbsnp_dir,
            set_pon=set_pon,
            pon_vcf=pon_vcf,
            cosmic_vcf=cosmic_vcf,
            cosmic_dir=cosmic_dir,
            tumor_name=sample_id + "_tumor",
            normal_name=sample_id + "_normal",
            docker=sentieon_docker,
            sample=sample_id,
            disk_size=disk_size,
            cluster_config=cluster_config
        }

        call annovar.annovar as TNseq_annovar {
            input:
            sample=sample_id,
            vcf=TNseq.TNseq_vcf,
            database=database,
            docker=annovar_docker,
            cluster_config=cluster_config,
            disk_size=disk_size
        }

        call vcf2maf.vcf2maf as TNseq_vcf2maf {
            input:
            sample=sample_id,
            multianno_txt=TNseq_annovar.multianno_txt,
            docker=maftools_docker,
            cluster_config=cluster_config,
            disk_size=disk_size
        }
    }

    if (tnscope) {
        call TNscope.TNscope as TNscope {
            input:
            SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
            SENTIEON_LICENSE=SENTIEON_LICENSE,
            fasta=fasta,
            ref_dir=ref_dir,
            regions=regions,
            corealigner_bam=corealigner.corealigner_bam,
            corealigner_bam_index=corealigner.corealigner_bam_index,
            dbsnp=dbsnp,
            dbsnp_dir=dbsnp_dir,
            set_pon=set_pon,
            pon_vcf=pon_vcf,
            cosmic_vcf=cosmic_vcf,
            cosmic_dir=cosmic_dir,
            tumor_name=sample_id + "_tumor",
            normal_name=sample_id + "_normal",
            docker=sentieon_docker,
            sample=sample_id,
            disk_size=disk_size,
            cluster_config=cluster_config
        }
        
        call annovar.annovar as TNscope_annovar {
            input:
            sample=sample_id,
            vcf=TNscope.TNscope_vcf,
            database=database,
            docker=annovar_docker,
            cluster_config=cluster_config,
            disk_size=disk_size
        }

        call vcf2maf.vcf2maf as TNscope_vcf2maf {
            input:
            sample=sample_id,
            multianno_txt=TNscope_annovar.multianno_txt,
            docker=maftools_docker,
            cluster_config=cluster_config,
            disk_size=disk_size
        }
    }

    if (varscan) {
        call somatic.somatic as somatic {
            input:
            sample=sample_id,
            tumor_bam=tumor_BQSR.recaled_bam,
            tumor_bam_index=tumor_BQSR.recaled_bam_index,
            normal_bam=normal_BQSR.recaled_bam,
            normal_bam_index=normal_BQSR.recaled_bam_index,
            ref_dir=ref_dir,
            fasta=fasta,        
            docker=varscan_docker,	
            disk_size=disk_size,
            cluster_config=cluster_config
        }

        call processSomatic.processSomatic as processSomatic {
            input:
            sample=sample_id,
            varscan_snp=somatic.varscan_snp,
            varscan_indel=somatic.varscan_indel,     
            docker=varscan_docker,	
            disk_size=disk_size,
            cluster_config=cluster_config
        }

        call somaticFilter.somaticFilter as somaticFilter {
            input:
            sample=sample_id,
            varscan_snp_somatic_hc=processSomatic.varscan_snp_somatic_hc,
            varscan_snp_germline_hc=processSomatic.varscan_snp_germline_hc,
            varscan_snp_loh_hc=processSomatic.varscan_snp_loh_hc,
            varscan_indel_somatic_hc=processSomatic.varscan_indel_somatic_hc,
            varscan_indel_germline_hc=processSomatic.varscan_indel_germline_hc,
            varscan_indel_loh_hc=processSomatic.varscan_indel_loh_hc,
            varscan_indel=somatic.varscan_indel,   
            docker=varscan_docker,	
            disk_size=disk_size,
            cluster_config=cluster_config
        }

        call annovar.annovar as snp_somatic_annovar {
            input: 
            sample=sample_id,
            vcf=somaticFilter.varscan_snp_somatic_filter,
            database=database,
            docker=annovar_docker,
            cluster_config=cluster_config,
            disk_size=disk_size
        }

        call vcf2maf.vcf2maf as snp_somatic_vcf2maf {
            input:
            sample=sample_id,
            multianno_txt=snp_somatic_annovar.multianno_txt,
            docker=maftools_docker,
            cluster_config=cluster_config,
            disk_size=disk_size
        }
        
        call annovar.annovar as snp_loh_annovar {
            input: 
            sample=sample_id,
            vcf=somaticFilter.varscan_snp_loh_filter,
            database=database,
            docker=annovar_docker,
            cluster_config=cluster_config,
            disk_size=disk_size
        }

        call vcf2maf.vcf2maf as snp_loh_vcf2maf {
            input:
            sample=sample_id,
            multianno_txt=snp_loh_annovar.multianno_txt,
            docker=maftools_docker,
            cluster_config=cluster_config,
            disk_size=disk_size
        }
        
        call annovar.annovar as indel_somatic_annovar {
            input: 
            sample=sample_id,
            vcf=somaticFilter.varscan_indel_somatic_filter,
            database=database,
            docker=annovar_docker,
            cluster_config=cluster_config,
            disk_size=disk_size
        }
        
        call vcf2maf.vcf2maf as indel_somatic_vcf2maf {
            input:
            sample=sample_id,
            multianno_txt=indel_somatic_annovar.multianno_txt,
            docker=maftools_docker,
            cluster_config=cluster_config,
            disk_size=disk_size
        }
        
        call annovar.annovar as indel_loh_annovar {
            input: 
            sample=sample_id,
            vcf=somaticFilter.varscan_indel_loh_filter,
            database=database,
            docker=annovar_docker,
            cluster_config=cluster_config,
            disk_size=disk_size
        }

        call vcf2maf.vcf2maf as indel_loh_vcf2maf {
            input:
            sample=sample_id,
            multianno_txt=indel_loh_annovar.multianno_txt,
            docker=maftools_docker,
            cluster_config=cluster_config,
            disk_size=disk_size
        }
    }
}